Adaptive Multiscale Computational Framework for Transient Problems

Rensselaer Polytechnic Institute, Troy NY

Investigators

Abstract

ABSTRACT The goal of the proposed research program is to develop an adaptive multiscale computational framework for transient problems aimed at predicting dynamic response of engineered components and structures including complex failure mechanisms operating at multiple temporal and spatial scales. The term multiscale computational framework is coined to emphasize that the dynamic behavior of structural systems is assessed from the fundamental physical processes operating at smaller spatial and temporal scales than currently resolved in simulations. The technical challenge is to develop new multiscale analysis and design concepts where material and structure are viewed as a single system and to enable the analyst to account in much more detail for the physics of the problem to ensure reliability of computations. Intellectual merit: The proposed framework will expand the fundamental computational and material mechanics knowledge in the following two respects: unified discrete-to-continuum and continuum-to-continuum scale bridging with concurrent consideration of multiple temporal and spatial scales, and adaptivity of hierarchical mathematical models. Broader impact: If successful, this effort will greatly impact science and industry's ability to model, analyze, and understand a vast array of multiscale systems in an accurate timely manner. Validation of these technologies will be performed on three feature applications: (a) Shock wave response of piezoelectric ceramics; (b) Energy absorption of honeycombs; (c) Crash prediction of polymer-based composite structural systems. These three feature applications have been carefully selected not only because they embody typical complexities associated with modeling multiple spatial and temporal scales, but more so due to their national economical impact (importance of energy absorbing material systems), and availability of experimental data critical for validation of computational capabilities. In addition to the three feature applications selected, other applications requiring large-scale, high-fidelity, predictive numerical simulation of naturally heterogeneous materials and engineered composites will benefit from this work. The technologies involved with this effort are enabling adaptive multiscale computational technologies associated with mechanical engineering, material science, physics, advanced modeling and software development. Educational materials including a new graduate course entitled Adaptive Multiscale Engineering Principles as well as tutorials and multiscale simulation tools will be developed.